It is common to have strategy parameters for different logic branches for a strategy. (i.e int parameter that 0 is off, 1 logic branch one, 2 logic branch two, etc)
Ideally it would be nice to have enum values that are optimizable so that when you see the optimization grid you see a readable value instead of just an int.
If the enum option isn't doable, it would be helpful to have a definable tooltip on the strategy parameter backtest/optimization screen that shows a tooltip when you hover over a given parameter. That way one can show "0 = off, 1 = logic two, etc"
Ideally it would be nice to have enum values that are optimizable so that when you see the optimization grid you see a readable value instead of just an int.
If the enum option isn't doable, it would be helpful to have a definable tooltip on the strategy parameter backtest/optimization screen that shows a tooltip when you hover over a given parameter. That way one can show "0 = off, 1 = logic two, etc"
Rename
QUOTE:
helpful to have a definable tooltip on the strategy parameter backtest/optimization screen that shows a tooltip when you hover over a given parameter.
That's already possible. It's called a "Hint".
CODE:
public VossPredictStrategy0() { AddParameter("Voss period", ParameterType.Int32, 22, 10, 28, 3) .Hint="Period for the Voss band-pass filter"; AddParameter("Voss predictor", ParameterType.Int32, 5, 2, 6, 1); AddParameter("Rank Confidence", ParameterType.Double, 0.60, 0.4, 0.9, 0.1) .Hint="Center position of the probably distribution to sample threshold candidates from"; AddParameter("DX period", ParameterType.Int32, 13, 5, 17, 4) .Hint="DSMA period for smoothing the DX time series"; }
Oh nice! Thanks superticker
There is a
CODE:that does exactly what you asked for.
ParameterType.StringChoice
But this can’t be optimized.
QUOTE:
But this can’t be optimized.
This is a pitty.
There are optimizer algorithms that distinguish quite well between continuous (double) discrete (int) and categorial (enum) variables.
Would be great if we could use this for the WL optimizers.
The existing optimizers could just cast the enum to an int (and back).
QUOTE:
The existing optimizers could just cast the enum to an int (and back).
Not so if the optimizer is using the derivative of the parameter to direct its rate of parameter change and direction. Many optimizers use a derivative to speed their convergence, and the parameters they are optimizing must be a "measure of distance".
But some optimizers are designed to work categorically and with a discontinuous function. Those should work for enum's.
Your Response
Post
Edit Post
Login is required